In [2]:
# Shane Khalid 10/17/2023
Import libraries. I normally use yfinance for this, but I have a clean dataset loading in¶
In [1]:
import numpy as np
import pandas as pd
import tensorflow as tf
import plotly.express as px
import statsmodels.api as sm
import matplotlib.pyplot as plt
import plotly.graph_objects as go
from tensorflow import keras
from keras.layers import GRU, Dropout, SimpleRNN, LSTM, Dense, SimpleRNN, GRU
from keras.models import Sequential
from sklearn.preprocessing import MinMaxScaler
print("Num GPUs Available: ", len(tf.config.experimental.list_physical_devices('GPU')))
2023-10-18 16:12:11.476119: E tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:9342] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered 2023-10-18 16:12:11.476178: E tensorflow/compiler/xla/stream_executor/cuda/cuda_fft.cc:609] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered 2023-10-18 16:12:11.477641: E tensorflow/compiler/xla/stream_executor/cuda/cuda_blas.cc:1518] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered 2023-10-18 16:12:11.610707: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations. To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
Num GPUs Available: 1
2023-10-18 16:12:15.660173: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:880] could not open file to read NUMA node: /sys/bus/pci/devices/0000:43:00.0/numa_node Your kernel may have been built without NUMA support. 2023-10-18 16:12:15.687115: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:880] could not open file to read NUMA node: /sys/bus/pci/devices/0000:43:00.0/numa_node Your kernel may have been built without NUMA support. 2023-10-18 16:12:15.687162: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:880] could not open file to read NUMA node: /sys/bus/pci/devices/0000:43:00.0/numa_node Your kernel may have been built without NUMA support.
Load Data (Google stock)¶
In [2]:
google_training_complete = pd.read_csv("./Google_Stock_Train (2010-2022).csv")
# Convert 'Date' column to datetime format
google_training_complete['Date'] = pd.to_datetime(google_training_complete['Date'])
google_training_complete.head(10)
Out[2]:
| Date | Open | High | Low | Close | Adj Close | Volume | |
|---|---|---|---|---|---|---|---|
| 0 | 2010-01-04 | 15.689439 | 15.753504 | 15.621622 | 15.684434 | 15.684434 | 78169752 |
| 1 | 2010-01-05 | 15.695195 | 15.711712 | 15.554054 | 15.615365 | 15.615365 | 120067812 |
| 2 | 2010-01-06 | 15.662162 | 15.662162 | 15.174174 | 15.221722 | 15.221722 | 158988852 |
| 3 | 2010-01-07 | 15.250250 | 15.265265 | 14.831081 | 14.867367 | 14.867367 | 256315428 |
| 4 | 2010-01-08 | 14.814815 | 15.096346 | 14.742492 | 15.065566 | 15.065566 | 188783028 |
| 5 | 2010-01-11 | 15.126627 | 15.126627 | 14.865866 | 15.042793 | 15.042793 | 288227484 |
| 6 | 2010-01-12 | 14.956206 | 14.968969 | 14.714715 | 14.776777 | 14.776777 | 193937868 |
| 7 | 2010-01-13 | 14.426677 | 14.724224 | 14.361862 | 14.691942 | 14.691942 | 259604136 |
| 8 | 2010-01-14 | 14.612112 | 14.869870 | 14.584835 | 14.761011 | 14.761011 | 169434396 |
| 9 | 2010-01-15 | 14.848348 | 14.853854 | 14.465465 | 14.514515 | 14.514515 | 217162620 |
Exploratory Data Analysis (EDA)¶
In [3]:
# Line chart for Google stock price over time
fig1 = px.line(google_training_complete, x='Date', y='Close', title='Google Stock Price Over Time')
fig1.show()
/home/shane/.local/lib/python3.10/site-packages/_plotly_utils/basevalidators.py:105: FutureWarning: The behavior of DatetimeProperties.to_pydatetime is deprecated, in a future version this will return a Series containing python datetime objects instead of an ndarray. To retain the old behavior, call `np.array` on the result v = v.dt.to_pydatetime()
In [4]:
# Scatter plot of daily trading volume
fig2 = px.scatter(google_training_complete, x='Date', y='Volume', title='Daily Trading Volume')
fig2.show()
/home/shane/.local/lib/python3.10/site-packages/_plotly_utils/basevalidators.py:105: FutureWarning: The behavior of DatetimeProperties.to_pydatetime is deprecated, in a future version this will return a Series containing python datetime objects instead of an ndarray. To retain the old behavior, call `np.array` on the result
In [5]:
# Box plot of Google stock prices for each year
google_training_complete['Year'] = google_training_complete['Date'].dt.year
fig3 = px.box(google_training_complete, x='Year', y='Close', title='Google Stock Prices - Yearly Box Plot')
fig3.show()
In [6]:
# Candlestick chart for stock prices
fig4 = go.Figure(data=[go.Candlestick(x=google_training_complete['Date'],
open=google_training_complete['Open'],
high=google_training_complete['High'],
low=google_training_complete['Low'],
close=google_training_complete['Close'])])
fig4.update_layout(title='Google Stock Prices - Candlestick Chart',
xaxis_title='Date',
yaxis_title='Stock Price')
fig4.show()
/home/shane/.local/lib/python3.10/site-packages/_plotly_utils/basevalidators.py:105: FutureWarning: The behavior of DatetimeProperties.to_pydatetime is deprecated, in a future version this will return a Series containing python datetime objects instead of an ndarray. To retain the old behavior, call `np.array` on the result
In [7]:
# Histogram of daily returns
google_training_complete['Daily_Return'] = google_training_complete['Close'].pct_change()
fig5 = px.histogram(google_training_complete, x='Daily_Return', nbins=30, title='Distribution of Daily Returns')
fig5.show()
In [9]:
# Heatmap of correlation matrix
correlation_matrix = google_training_complete.corr()
fig6 = px.imshow(correlation_matrix, x=correlation_matrix.index, y=correlation_matrix.columns, title='Correlation Matrix Heatmap')
fig6.show()
In [10]:
# Moving Average of closing prices
google_training_complete['MA_50'] = google_training_complete['Close'].rolling(window=50).mean()
fig7 = px.line(google_training_complete, x='Date', y=['Close', 'MA_50'], title='Google Stock Close Price with 50-Day Moving Average')
fig7.show()
/home/shane/.local/lib/python3.10/site-packages/_plotly_utils/basevalidators.py:105: FutureWarning: The behavior of DatetimeProperties.to_pydatetime is deprecated, in a future version this will return a Series containing python datetime objects instead of an ndarray. To retain the old behavior, call `np.array` on the result
In [11]:
# Scatter plot of closing prices vs. trading volume
fig8 = px.scatter(google_training_complete, x='Close', y='Volume', title='Closing Prices vs. Trading Volume')
fig8.show()
In [12]:
# Line chart for daily stock price change
google_training_complete['Daily_Change'] = google_training_complete['Close'].diff()
fig9 = px.line(google_training_complete, x='Date', y='Daily_Change', title='Daily Stock Price Change')
fig9.show()
/home/shane/.local/lib/python3.10/site-packages/_plotly_utils/basevalidators.py:105: FutureWarning: The behavior of DatetimeProperties.to_pydatetime is deprecated, in a future version this will return a Series containing python datetime objects instead of an ndarray. To retain the old behavior, call `np.array` on the result
In [13]:
# Bar chart of trading volume per month
google_training_complete['Month'] = google_training_complete['Date'].dt.month
monthly_volume = google_training_complete.groupby('Month')['Volume'].sum().reset_index()
fig10 = px.bar(monthly_volume, x='Month', y='Volume', title='Total Trading Volume per Month')
fig10.show()
In [14]:
# Area chart for daily trading volume
fig11 = px.area(google_training_complete, x='Date', y='Volume', title='Daily Trading Volume (Area Chart)')
fig11.show()
/home/shane/.local/lib/python3.10/site-packages/_plotly_utils/basevalidators.py:105: FutureWarning: The behavior of DatetimeProperties.to_pydatetime is deprecated, in a future version this will return a Series containing python datetime objects instead of an ndarray. To retain the old behavior, call `np.array` on the result
In [15]:
# Histogram of daily stock price fluctuations
fig12 = px.histogram(google_training_complete, x='Daily_Change', title='Distribution of Daily Stock Price Changes')
fig12.show()
In [16]:
# Line chart for percent change of daily stock price
google_training_complete['Daily_Pct_Change'] = google_training_complete['Close'].pct_change() * 100
fig13 = px.line(google_training_complete, x='Date', y='Daily_Pct_Change', title='Daily Stock Price Percent Change')
fig13.show()
/home/shane/.local/lib/python3.10/site-packages/_plotly_utils/basevalidators.py:105: FutureWarning: The behavior of DatetimeProperties.to_pydatetime is deprecated, in a future version this will return a Series containing python datetime objects instead of an ndarray. To retain the old behavior, call `np.array` on the result
In [17]:
# Box plot of daily stock price percent change by month
fig14 = px.box(google_training_complete, x='Month', y='Daily_Pct_Change', title='Daily Stock Price Percent Change by Month')
fig14.show()
In [18]:
# Scatter plot of closing prices with Trend Line (should have used differnt color)
fig15 = px.scatter(google_training_complete, x='Date', y='Close', title='Google Stock Prices with Trendline', trendline='lowess')
fig15.show()
/home/shane/.local/lib/python3.10/site-packages/_plotly_utils/basevalidators.py:105: FutureWarning: The behavior of DatetimeProperties.to_pydatetime is deprecated, in a future version this will return a Series containing python datetime objects instead of an ndarray. To retain the old behavior, call `np.array` on the result /home/shane/.local/lib/python3.10/site-packages/_plotly_utils/basevalidators.py:105: FutureWarning: The behavior of DatetimeProperties.to_pydatetime is deprecated, in a future version this will return a Series containing python datetime objects instead of an ndarray. To retain the old behavior, call `np.array` on the result
In [19]:
# Line chart for daily trading volume with moving average
google_training_complete['MA_Volume'] = google_training_complete['Volume'].rolling(window=10).mean()
fig16 = px.line(google_training_complete, x='Date', y=['Volume', 'MA_Volume'], title='Daily Trading Volume with 10-Day Moving Average')
fig16.show()
/home/shane/.local/lib/python3.10/site-packages/_plotly_utils/basevalidators.py:105: FutureWarning: The behavior of DatetimeProperties.to_pydatetime is deprecated, in a future version this will return a Series containing python datetime objects instead of an ndarray. To retain the old behavior, call `np.array` on the result
In [20]:
# Box plot of daily stock price changes by year
google_training_complete['Year'] = google_training_complete['Date'].dt.year
fig17 = px.box(google_training_complete, x='Year', y='Daily_Change', title='Daily Stock Price Changes by Year')
fig17.show()
In [21]:
# Line chart for daily closing prices in 2022
# 2022 because Train dataset is 2010-2022
df_2022 = google_training_complete[google_training_complete['Year'] == 2022]
fig18 = px.line(google_training_complete, x='Date', y='Close', title='Google Stock Prices in 2022')
fig18.show()
/home/shane/.local/lib/python3.10/site-packages/_plotly_utils/basevalidators.py:105: FutureWarning: The behavior of DatetimeProperties.to_pydatetime is deprecated, in a future version this will return a Series containing python datetime objects instead of an ndarray. To retain the old behavior, call `np.array` on the result
In [22]:
# Scatter plot of daily returns vs. trading volume
fig19 = px.scatter(google_training_complete, x='Daily_Return', y='Volume', title='Daily Returns vs. Trading Volume')
fig19.show()
In [23]:
# Line chart for daily trading volume with range slider
fig20 = px.line(google_training_complete, x='Date', y='Volume', title='Daily Trading Volume with Range Slider')
fig20.update_xaxes(rangeslider_visible=True)
fig20.show()
/home/shane/.local/lib/python3.10/site-packages/_plotly_utils/basevalidators.py:105: FutureWarning: The behavior of DatetimeProperties.to_pydatetime is deprecated, in a future version this will return a Series containing python datetime objects instead of an ndarray. To retain the old behavior, call `np.array` on the result
Data preprocessing¶
In [24]:
google_training_processed = google_training_complete.iloc[:, 4:5].values
google_training_processed
Out[24]:
array([[15.684434],
[15.615365],
[15.221722],
...,
[86.019997],
[88.449997],
[88.230003]])
In [25]:
scaler = MinMaxScaler(feature_range = (0, 1))
google_training_scaled = scaler.fit_transform(google_training_processed)
In [26]:
print(google_training_scaled)
print(google_training_scaled.shape)
[[0.03434761] [0.03385045] [0.03101697] ... [0.54062898] [0.55812033] [0.55653679]] (3272, 1)
In [27]:
features_set = []
labels = []
for i in range(60, google_training_scaled.shape[0]):
features_set.append(google_training_scaled[i-60:i, 0])
labels.append(google_training_scaled[i, 0])
In [28]:
type(features_set)
Out[28]:
list
In [29]:
features_set, labels = np.array(features_set), np.array(labels)
In [30]:
features_set = np.reshape(features_set, (features_set.shape[0], features_set.shape[1], 1))
print(features_set.shape)
(3212, 60, 1)
LSTM layers work on 3D data with the following structure (nb_sequence, nb_timestep, nb_feature).¶
- nb_sequence : total number of sequences in dataset
- nb_timestep : size of sequences
- nb_feature : number of features describing timesteps RNN¶
In [31]:
RNN_model = Sequential()
RNN_model.add(SimpleRNN(units=300, return_sequences=True, input_shape=(features_set.shape[1], 1)))
RNN_model.add(Dropout(0.2))
RNN_model.add(SimpleRNN(units=100, return_sequences=True))
RNN_model.add(Dropout(0.2))
RNN_model.add(SimpleRNN(units=100, return_sequences=True))
RNN_model.add(Dropout(0.2))
RNN_model.add(SimpleRNN(units=100))
RNN_model.add(Dropout(0.2))
RNN_model.add(Dense(units = 1))
2023-10-18 16:14:36.082137: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:880] could not open file to read NUMA node: /sys/bus/pci/devices/0000:43:00.0/numa_node Your kernel may have been built without NUMA support. 2023-10-18 16:14:36.082241: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:880] could not open file to read NUMA node: /sys/bus/pci/devices/0000:43:00.0/numa_node Your kernel may have been built without NUMA support. 2023-10-18 16:14:36.082271: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:880] could not open file to read NUMA node: /sys/bus/pci/devices/0000:43:00.0/numa_node Your kernel may have been built without NUMA support. 2023-10-18 16:14:36.468842: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:880] could not open file to read NUMA node: /sys/bus/pci/devices/0000:43:00.0/numa_node Your kernel may have been built without NUMA support. 2023-10-18 16:14:36.468912: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:880] could not open file to read NUMA node: /sys/bus/pci/devices/0000:43:00.0/numa_node Your kernel may have been built without NUMA support. 2023-10-18 16:14:36.468924: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1977] Could not identify NUMA node of platform GPU id 0, defaulting to 0. Your kernel may not have been built with NUMA support. 2023-10-18 16:14:36.468972: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:880] could not open file to read NUMA node: /sys/bus/pci/devices/0000:43:00.0/numa_node Your kernel may have been built without NUMA support. 2023-10-18 16:14:36.469000: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1886] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 21765 MB memory: -> device: 0, name: NVIDIA GeForce RTX 3090 Ti, pci bus id: 0000:43:00.0, compute capability: 8.6 2023-10-18 16:14:40.842066: I tensorflow/tsl/platform/default/subprocess.cc:304] Start cannot spawn child process: No such file or directory
Compile RNN Model using Adam optimizer¶
In [32]:
RNN_model.compile(optimizer = 'adam', loss = 'mean_squared_error')
In [33]:
RNN_model.summary()
Model: "sequential"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
simple_rnn (SimpleRNN) (None, 60, 300) 90600
dropout (Dropout) (None, 60, 300) 0
simple_rnn_1 (SimpleRNN) (None, 60, 100) 40100
dropout_1 (Dropout) (None, 60, 100) 0
simple_rnn_2 (SimpleRNN) (None, 60, 100) 20100
dropout_2 (Dropout) (None, 60, 100) 0
simple_rnn_3 (SimpleRNN) (None, 100) 20100
dropout_3 (Dropout) (None, 100) 0
dense (Dense) (None, 1) 101
=================================================================
Total params: 171001 (667.97 KB)
Trainable params: 171001 (667.97 KB)
Non-trainable params: 0 (0.00 Byte)
_________________________________________________________________
Fit RNN Model¶
In [40]:
# Finally got the GPU backend to work
# batch sizes of 32 led to a capped 24% GPU usage but raising it to 3200 made the GPU usage >50% and it proccessed the epochs lightning fast
RNN_History = RNN_model.fit(features_set, labels, epochs = 300, batch_size = 3200)
Epoch 1/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0494 Epoch 2/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0424 Epoch 3/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0404 Epoch 4/300 2/2 [==============================] - 1s 382ms/step - loss: 0.0321 Epoch 5/300 2/2 [==============================] - 1s 398ms/step - loss: 0.0377 Epoch 6/300 2/2 [==============================] - 1s 389ms/step - loss: 0.0249 Epoch 7/300 2/2 [==============================] - 1s 397ms/step - loss: 0.0131 Epoch 8/300 2/2 [==============================] - 1s 356ms/step - loss: 0.0131 Epoch 9/300 2/2 [==============================] - 1s 361ms/step - loss: 0.0112 Epoch 10/300 2/2 [==============================] - 1s 351ms/step - loss: 0.0152 Epoch 11/300 2/2 [==============================] - 1s 349ms/step - loss: 0.0427 Epoch 12/300 2/2 [==============================] - 1s 337ms/step - loss: 0.0515 Epoch 13/300 2/2 [==============================] - 1s 343ms/step - loss: 0.0295 Epoch 14/300 2/2 [==============================] - 1s 363ms/step - loss: 0.0655 Epoch 15/300 2/2 [==============================] - 1s 379ms/step - loss: 0.0370 Epoch 16/300 2/2 [==============================] - 1s 368ms/step - loss: 0.0411 Epoch 17/300 2/2 [==============================] - 1s 343ms/step - loss: 0.0180 Epoch 18/300 2/2 [==============================] - 1s 339ms/step - loss: 0.0119 Epoch 19/300 2/2 [==============================] - 1s 330ms/step - loss: 0.0128 Epoch 20/300 2/2 [==============================] - 1s 347ms/step - loss: 0.0323 Epoch 21/300 2/2 [==============================] - 1s 349ms/step - loss: 0.0174 Epoch 22/300 2/2 [==============================] - 1s 356ms/step - loss: 0.0158 Epoch 23/300 2/2 [==============================] - 1s 341ms/step - loss: 0.0116 Epoch 24/300 2/2 [==============================] - 1s 341ms/step - loss: 0.0160 Epoch 25/300 2/2 [==============================] - 1s 345ms/step - loss: 0.0121 Epoch 26/300 2/2 [==============================] - 1s 357ms/step - loss: 0.0118 Epoch 27/300 2/2 [==============================] - 1s 354ms/step - loss: 0.0088 Epoch 28/300 2/2 [==============================] - 1s 358ms/step - loss: 0.0080 Epoch 29/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0084 Epoch 30/300 2/2 [==============================] - 1s 358ms/step - loss: 0.0132 Epoch 31/300 2/2 [==============================] - 1s 342ms/step - loss: 0.0290 Epoch 32/300 2/2 [==============================] - 1s 376ms/step - loss: 0.0087 Epoch 33/300 2/2 [==============================] - 1s 367ms/step - loss: 0.0082 Epoch 34/300 2/2 [==============================] - 1s 361ms/step - loss: 0.0092 Epoch 35/300 2/2 [==============================] - 1s 409ms/step - loss: 0.0095 Epoch 36/300 2/2 [==============================] - 1s 342ms/step - loss: 0.0095 Epoch 37/300 2/2 [==============================] - 1s 351ms/step - loss: 0.0086 Epoch 38/300 2/2 [==============================] - 1s 360ms/step - loss: 0.0091 Epoch 39/300 2/2 [==============================] - 1s 356ms/step - loss: 0.0120 Epoch 40/300 2/2 [==============================] - 1s 471ms/step - loss: 0.0129 Epoch 41/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0102 Epoch 42/300 2/2 [==============================] - 1s 346ms/step - loss: 0.0161 Epoch 43/300 2/2 [==============================] - 1s 377ms/step - loss: 0.0084 Epoch 44/300 2/2 [==============================] - 1s 388ms/step - loss: 0.0102 Epoch 45/300 2/2 [==============================] - 1s 349ms/step - loss: 0.0176 Epoch 46/300 2/2 [==============================] - 1s 345ms/step - loss: 0.0232 Epoch 47/300 2/2 [==============================] - 1s 349ms/step - loss: 0.0106 Epoch 48/300 2/2 [==============================] - 1s 354ms/step - loss: 0.0133 Epoch 49/300 2/2 [==============================] - 1s 365ms/step - loss: 0.0066 Epoch 50/300 2/2 [==============================] - 1s 360ms/step - loss: 0.0069 Epoch 51/300 2/2 [==============================] - 1s 352ms/step - loss: 0.0092 Epoch 52/300 2/2 [==============================] - 1s 336ms/step - loss: 0.0135 Epoch 53/300 2/2 [==============================] - 1s 341ms/step - loss: 0.0085 Epoch 54/300 2/2 [==============================] - 1s 335ms/step - loss: 0.0142 Epoch 55/300 2/2 [==============================] - 1s 343ms/step - loss: 0.0080 Epoch 56/300 2/2 [==============================] - 1s 343ms/step - loss: 0.0090 Epoch 57/300 2/2 [==============================] - 1s 350ms/step - loss: 0.0108 Epoch 58/300 2/2 [==============================] - 1s 333ms/step - loss: 0.0069 Epoch 59/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0133 Epoch 60/300 2/2 [==============================] - 1s 337ms/step - loss: 0.0061 Epoch 61/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0063 Epoch 62/300 2/2 [==============================] - 1s 365ms/step - loss: 0.0089 Epoch 63/300 2/2 [==============================] - 1s 347ms/step - loss: 0.0054 Epoch 64/300 2/2 [==============================] - 1s 352ms/step - loss: 0.0072 Epoch 65/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0066 Epoch 66/300 2/2 [==============================] - 1s 331ms/step - loss: 0.0064 Epoch 67/300 2/2 [==============================] - 1s 346ms/step - loss: 0.0086 Epoch 68/300 2/2 [==============================] - 1s 336ms/step - loss: 0.0052 Epoch 69/300 2/2 [==============================] - 1s 338ms/step - loss: 0.0072 Epoch 70/300 2/2 [==============================] - 1s 344ms/step - loss: 0.0086 Epoch 71/300 2/2 [==============================] - 1s 370ms/step - loss: 0.0054 Epoch 72/300 2/2 [==============================] - 1s 355ms/step - loss: 0.0061 Epoch 73/300 2/2 [==============================] - 1s 347ms/step - loss: 0.0062 Epoch 74/300 2/2 [==============================] - 1s 350ms/step - loss: 0.0052 Epoch 75/300 2/2 [==============================] - 1s 346ms/step - loss: 0.0054 Epoch 76/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0054 Epoch 77/300 2/2 [==============================] - 1s 357ms/step - loss: 0.0051 Epoch 78/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0070 Epoch 79/300 2/2 [==============================] - 1s 347ms/step - loss: 0.0046 Epoch 80/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0050 Epoch 81/300 2/2 [==============================] - 1s 344ms/step - loss: 0.0052 Epoch 82/300 2/2 [==============================] - 1s 341ms/step - loss: 0.0050 Epoch 83/300 2/2 [==============================] - 1s 362ms/step - loss: 0.0052 Epoch 84/300 2/2 [==============================] - 1s 342ms/step - loss: 0.0054 Epoch 85/300 2/2 [==============================] - 1s 363ms/step - loss: 0.0049 Epoch 86/300 2/2 [==============================] - 1s 343ms/step - loss: 0.0053 Epoch 87/300 2/2 [==============================] - 1s 341ms/step - loss: 0.0045 Epoch 88/300 2/2 [==============================] - 1s 349ms/step - loss: 0.0054 Epoch 89/300 2/2 [==============================] - 1s 346ms/step - loss: 0.0054 Epoch 90/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0045 Epoch 91/300 2/2 [==============================] - 1s 351ms/step - loss: 0.0060 Epoch 92/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0064 Epoch 93/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0044 Epoch 94/300 2/2 [==============================] - 1s 365ms/step - loss: 0.0061 Epoch 95/300 2/2 [==============================] - 1s 340ms/step - loss: 0.0049 Epoch 96/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0043 Epoch 97/300 2/2 [==============================] - 1s 356ms/step - loss: 0.0042 Epoch 98/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0046 Epoch 99/300 2/2 [==============================] - 1s 367ms/step - loss: 0.0041 Epoch 100/300 2/2 [==============================] - 1s 360ms/step - loss: 0.0045 Epoch 101/300 2/2 [==============================] - 1s 350ms/step - loss: 0.0050 Epoch 102/300 2/2 [==============================] - 1s 349ms/step - loss: 0.0046 Epoch 103/300 2/2 [==============================] - 1s 346ms/step - loss: 0.0043 Epoch 104/300 2/2 [==============================] - 1s 391ms/step - loss: 0.0049 Epoch 105/300 2/2 [==============================] - 1s 358ms/step - loss: 0.0050 Epoch 106/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0041 Epoch 107/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0058 Epoch 108/300 2/2 [==============================] - 1s 346ms/step - loss: 0.0062 Epoch 109/300 2/2 [==============================] - 1s 342ms/step - loss: 0.0040 Epoch 110/300 2/2 [==============================] - 1s 347ms/step - loss: 0.0049 Epoch 111/300 2/2 [==============================] - 1s 344ms/step - loss: 0.0070 Epoch 112/300 2/2 [==============================] - 1s 368ms/step - loss: 0.0046 Epoch 113/300 2/2 [==============================] - 1s 355ms/step - loss: 0.0052 Epoch 114/300 2/2 [==============================] - 1s 356ms/step - loss: 0.0046 Epoch 115/300 2/2 [==============================] - 1s 357ms/step - loss: 0.0043 Epoch 116/300 2/2 [==============================] - 1s 339ms/step - loss: 0.0051 Epoch 117/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0051 Epoch 118/300 2/2 [==============================] - 1s 363ms/step - loss: 0.0045 Epoch 119/300 2/2 [==============================] - 1s 408ms/step - loss: 0.0056 Epoch 120/300 2/2 [==============================] - 1s 363ms/step - loss: 0.0065 Epoch 121/300 2/2 [==============================] - 1s 352ms/step - loss: 0.0062 Epoch 122/300 2/2 [==============================] - 1s 346ms/step - loss: 0.0071 Epoch 123/300 2/2 [==============================] - 1s 341ms/step - loss: 0.0042 Epoch 124/300 2/2 [==============================] - 1s 347ms/step - loss: 0.0085 Epoch 125/300 2/2 [==============================] - 1s 346ms/step - loss: 0.0043 Epoch 126/300 2/2 [==============================] - 1s 350ms/step - loss: 0.0069 Epoch 127/300 2/2 [==============================] - 1s 354ms/step - loss: 0.0052 Epoch 128/300 2/2 [==============================] - 1s 347ms/step - loss: 0.0047 Epoch 129/300 2/2 [==============================] - 1s 356ms/step - loss: 0.0068 Epoch 130/300 2/2 [==============================] - 1s 345ms/step - loss: 0.0036 Epoch 131/300 2/2 [==============================] - 1s 355ms/step - loss: 0.0049 Epoch 132/300 2/2 [==============================] - 1s 360ms/step - loss: 0.0056 Epoch 133/300 2/2 [==============================] - 1s 354ms/step - loss: 0.0045 Epoch 134/300 2/2 [==============================] - 1s 352ms/step - loss: 0.0045 Epoch 135/300 2/2 [==============================] - 1s 408ms/step - loss: 0.0056 Epoch 136/300 2/2 [==============================] - 1s 359ms/step - loss: 0.0040 Epoch 137/300 2/2 [==============================] - 1s 331ms/step - loss: 0.0064 Epoch 138/300 2/2 [==============================] - 1s 347ms/step - loss: 0.0046 Epoch 139/300 2/2 [==============================] - 1s 337ms/step - loss: 0.0051 Epoch 140/300 2/2 [==============================] - 1s 367ms/step - loss: 0.0046 Epoch 141/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0046 Epoch 142/300 2/2 [==============================] - 1s 340ms/step - loss: 0.0046 Epoch 143/300 2/2 [==============================] - 1s 365ms/step - loss: 0.0078 Epoch 144/300 2/2 [==============================] - 1s 346ms/step - loss: 0.0048 Epoch 145/300 2/2 [==============================] - 1s 350ms/step - loss: 0.0038 Epoch 146/300 2/2 [==============================] - 1s 340ms/step - loss: 0.0037 Epoch 147/300 2/2 [==============================] - 1s 346ms/step - loss: 0.0044 Epoch 148/300 2/2 [==============================] - 1s 333ms/step - loss: 0.0046 Epoch 149/300 2/2 [==============================] - 1s 354ms/step - loss: 0.0046 Epoch 150/300 2/2 [==============================] - 1s 352ms/step - loss: 0.0039 Epoch 151/300 2/2 [==============================] - 1s 345ms/step - loss: 0.0040 Epoch 152/300 2/2 [==============================] - 1s 361ms/step - loss: 0.0049 Epoch 153/300 2/2 [==============================] - 1s 334ms/step - loss: 0.0047 Epoch 154/300 2/2 [==============================] - 1s 362ms/step - loss: 0.0038 Epoch 155/300 2/2 [==============================] - 1s 408ms/step - loss: 0.0067 Epoch 156/300 2/2 [==============================] - 1s 338ms/step - loss: 0.0038 Epoch 157/300 2/2 [==============================] - 1s 364ms/step - loss: 0.0044 Epoch 158/300 2/2 [==============================] - 1s 351ms/step - loss: 0.0039 Epoch 159/300 2/2 [==============================] - 1s 332ms/step - loss: 0.0050 Epoch 160/300 2/2 [==============================] - 1s 349ms/step - loss: 0.0042 Epoch 161/300 2/2 [==============================] - 1s 386ms/step - loss: 0.0041 Epoch 162/300 2/2 [==============================] - 1s 357ms/step - loss: 0.0046 Epoch 163/300 2/2 [==============================] - 1s 335ms/step - loss: 0.0038 Epoch 164/300 2/2 [==============================] - 1s 340ms/step - loss: 0.0050 Epoch 165/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0039 Epoch 166/300 2/2 [==============================] - 1s 347ms/step - loss: 0.0042 Epoch 167/300 2/2 [==============================] - 1s 336ms/step - loss: 0.0058 Epoch 168/300 2/2 [==============================] - 1s 340ms/step - loss: 0.0038 Epoch 169/300 2/2 [==============================] - 1s 343ms/step - loss: 0.0050 Epoch 170/300 2/2 [==============================] - 1s 338ms/step - loss: 0.0046 Epoch 171/300 2/2 [==============================] - 1s 344ms/step - loss: 0.0056 Epoch 172/300 2/2 [==============================] - 1s 343ms/step - loss: 0.0040 Epoch 173/300 2/2 [==============================] - 1s 334ms/step - loss: 0.0034 Epoch 174/300 2/2 [==============================] - 1s 339ms/step - loss: 0.0040 Epoch 175/300 2/2 [==============================] - 1s 344ms/step - loss: 0.0038 Epoch 176/300 2/2 [==============================] - 1s 347ms/step - loss: 0.0035 Epoch 177/300 2/2 [==============================] - 1s 343ms/step - loss: 0.0033 Epoch 178/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0037 Epoch 179/300 2/2 [==============================] - 1s 441ms/step - loss: 0.0037 Epoch 180/300 2/2 [==============================] - 1s 336ms/step - loss: 0.0032 Epoch 181/300 2/2 [==============================] - 1s 408ms/step - loss: 0.0038 Epoch 182/300 2/2 [==============================] - 1s 366ms/step - loss: 0.0035 Epoch 183/300 2/2 [==============================] - 1s 351ms/step - loss: 0.0033 Epoch 184/300 2/2 [==============================] - 1s 351ms/step - loss: 0.0035 Epoch 185/300 2/2 [==============================] - 1s 357ms/step - loss: 0.0033 Epoch 186/300 2/2 [==============================] - 1s 380ms/step - loss: 0.0039 Epoch 187/300 2/2 [==============================] - 1s 337ms/step - loss: 0.0039 Epoch 188/300 2/2 [==============================] - 1s 346ms/step - loss: 0.0039 Epoch 189/300 2/2 [==============================] - 1s 349ms/step - loss: 0.0037 Epoch 190/300 2/2 [==============================] - 1s 346ms/step - loss: 0.0050 Epoch 191/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0037 Epoch 192/300 2/2 [==============================] - 1s 352ms/step - loss: 0.0035 Epoch 193/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0040 Epoch 194/300 2/2 [==============================] - 1s 349ms/step - loss: 0.0036 Epoch 195/300 2/2 [==============================] - 1s 368ms/step - loss: 0.0036 Epoch 196/300 2/2 [==============================] - 1s 363ms/step - loss: 0.0037 Epoch 197/300 2/2 [==============================] - 1s 344ms/step - loss: 0.0046 Epoch 198/300 2/2 [==============================] - 1s 352ms/step - loss: 0.0034 Epoch 199/300 2/2 [==============================] - 1s 351ms/step - loss: 0.0033 Epoch 200/300 2/2 [==============================] - 1s 356ms/step - loss: 0.0037 Epoch 201/300 2/2 [==============================] - 1s 345ms/step - loss: 0.0032 Epoch 202/300 2/2 [==============================] - 1s 339ms/step - loss: 0.0032 Epoch 203/300 2/2 [==============================] - 1s 356ms/step - loss: 0.0040 Epoch 204/300 2/2 [==============================] - 1s 331ms/step - loss: 0.0035 Epoch 205/300 2/2 [==============================] - 1s 343ms/step - loss: 0.0032 Epoch 206/300 2/2 [==============================] - 1s 346ms/step - loss: 0.0031 Epoch 207/300 2/2 [==============================] - 1s 355ms/step - loss: 0.0033 Epoch 208/300 2/2 [==============================] - 1s 360ms/step - loss: 0.0032 Epoch 209/300 2/2 [==============================] - 1s 350ms/step - loss: 0.0035 Epoch 210/300 2/2 [==============================] - 1s 334ms/step - loss: 0.0033 Epoch 211/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0032 Epoch 212/300 2/2 [==============================] - 1s 346ms/step - loss: 0.0034 Epoch 213/300 2/2 [==============================] - 1s 357ms/step - loss: 0.0040 Epoch 214/300 2/2 [==============================] - 1s 345ms/step - loss: 0.0034 Epoch 215/300 2/2 [==============================] - 1s 390ms/step - loss: 0.0036 Epoch 216/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0041 Epoch 217/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0035 Epoch 218/300 2/2 [==============================] - 1s 345ms/step - loss: 0.0038 Epoch 219/300 2/2 [==============================] - 1s 370ms/step - loss: 0.0040 Epoch 220/300 2/2 [==============================] - 1s 338ms/step - loss: 0.0036 Epoch 221/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0038 Epoch 222/300 2/2 [==============================] - 1s 342ms/step - loss: 0.0041 Epoch 223/300 2/2 [==============================] - 1s 341ms/step - loss: 0.0033 Epoch 224/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0037 Epoch 225/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0038 Epoch 226/300 2/2 [==============================] - 1s 349ms/step - loss: 0.0034 Epoch 227/300 2/2 [==============================] - 1s 350ms/step - loss: 0.0042 Epoch 228/300 2/2 [==============================] - 1s 345ms/step - loss: 0.0041 Epoch 229/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0032 Epoch 230/300 2/2 [==============================] - 1s 349ms/step - loss: 0.0045 Epoch 231/300 2/2 [==============================] - 1s 349ms/step - loss: 0.0048 Epoch 232/300 2/2 [==============================] - 1s 355ms/step - loss: 0.0033 Epoch 233/300 2/2 [==============================] - 1s 354ms/step - loss: 0.0052 Epoch 234/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0034 Epoch 235/300 2/2 [==============================] - 1s 347ms/step - loss: 0.0056 Epoch 236/300 2/2 [==============================] - 1s 347ms/step - loss: 0.0033 Epoch 237/300 2/2 [==============================] - 1s 337ms/step - loss: 0.0041 Epoch 238/300 2/2 [==============================] - 1s 344ms/step - loss: 0.0033 Epoch 239/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0040 Epoch 240/300 2/2 [==============================] - 1s 359ms/step - loss: 0.0031 Epoch 241/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0040 Epoch 242/300 2/2 [==============================] - 1s 331ms/step - loss: 0.0033 Epoch 243/300 2/2 [==============================] - 1s 350ms/step - loss: 0.0032 Epoch 244/300 2/2 [==============================] - 1s 350ms/step - loss: 0.0032 Epoch 245/300 2/2 [==============================] - 1s 352ms/step - loss: 0.0032 Epoch 246/300 2/2 [==============================] - 1s 340ms/step - loss: 0.0030 Epoch 247/300 2/2 [==============================] - 1s 352ms/step - loss: 0.0032 Epoch 248/300 2/2 [==============================] - 1s 339ms/step - loss: 0.0032 Epoch 249/300 2/2 [==============================] - 1s 341ms/step - loss: 0.0033 Epoch 250/300 2/2 [==============================] - 1s 345ms/step - loss: 0.0032 Epoch 251/300 2/2 [==============================] - 1s 364ms/step - loss: 0.0033 Epoch 252/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0029 Epoch 253/300 2/2 [==============================] - 1s 347ms/step - loss: 0.0032 Epoch 254/300 2/2 [==============================] - 1s 345ms/step - loss: 0.0035 Epoch 255/300 2/2 [==============================] - 1s 374ms/step - loss: 0.0033 Epoch 256/300 2/2 [==============================] - 1s 337ms/step - loss: 0.0041 Epoch 257/300 2/2 [==============================] - 1s 369ms/step - loss: 0.0031 Epoch 258/300 2/2 [==============================] - 1s 351ms/step - loss: 0.0033 Epoch 259/300 2/2 [==============================] - 1s 354ms/step - loss: 0.0046 Epoch 260/300 2/2 [==============================] - 1s 353ms/step - loss: 0.0043 Epoch 261/300 2/2 [==============================] - 1s 350ms/step - loss: 0.0039 Epoch 262/300 2/2 [==============================] - 1s 354ms/step - loss: 0.0031 Epoch 263/300 2/2 [==============================] - 1s 339ms/step - loss: 0.0031 Epoch 264/300 2/2 [==============================] - 1s 339ms/step - loss: 0.0076 Epoch 265/300 2/2 [==============================] - 1s 356ms/step - loss: 0.0032 Epoch 266/300 2/2 [==============================] - 1s 358ms/step - loss: 0.0077 Epoch 267/300 2/2 [==============================] - 1s 361ms/step - loss: 0.0044 Epoch 268/300 2/2 [==============================] - 1s 344ms/step - loss: 0.0041 Epoch 269/300 2/2 [==============================] - 1s 384ms/step - loss: 0.0063 Epoch 270/300 2/2 [==============================] - 1s 389ms/step - loss: 0.0033 Epoch 271/300 2/2 [==============================] - 1s 352ms/step - loss: 0.0051 Epoch 272/300 2/2 [==============================] - 1s 356ms/step - loss: 0.0033 Epoch 273/300 2/2 [==============================] - 1s 349ms/step - loss: 0.0034 Epoch 274/300 2/2 [==============================] - 1s 385ms/step - loss: 0.0030 Epoch 275/300 2/2 [==============================] - 1s 336ms/step - loss: 0.0033 Epoch 276/300 2/2 [==============================] - 1s 358ms/step - loss: 0.0031 Epoch 277/300 2/2 [==============================] - 1s 362ms/step - loss: 0.0039 Epoch 278/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0031 Epoch 279/300 2/2 [==============================] - 1s 357ms/step - loss: 0.0035 Epoch 280/300 2/2 [==============================] - 1s 331ms/step - loss: 0.0077 Epoch 281/300 2/2 [==============================] - 1s 378ms/step - loss: 0.0029 Epoch 282/300 2/2 [==============================] - 1s 378ms/step - loss: 0.0042 Epoch 283/300 2/2 [==============================] - 1s 360ms/step - loss: 0.0050 Epoch 284/300 2/2 [==============================] - 1s 348ms/step - loss: 0.0033 Epoch 285/300 2/2 [==============================] - 1s 346ms/step - loss: 0.0031 Epoch 286/300 2/2 [==============================] - 1s 346ms/step - loss: 0.0040 Epoch 287/300 2/2 [==============================] - 1s 354ms/step - loss: 0.0035 Epoch 288/300 2/2 [==============================] - 1s 368ms/step - loss: 0.0045 Epoch 289/300 2/2 [==============================] - 1s 359ms/step - loss: 0.0032 Epoch 290/300 2/2 [==============================] - 1s 372ms/step - loss: 0.0045 Epoch 291/300 2/2 [==============================] - 1s 369ms/step - loss: 0.0033 Epoch 292/300 2/2 [==============================] - 1s 361ms/step - loss: 0.0030 Epoch 293/300 2/2 [==============================] - 1s 403ms/step - loss: 0.0041 Epoch 294/300 2/2 [==============================] - 1s 362ms/step - loss: 0.0034 Epoch 295/300 2/2 [==============================] - 1s 337ms/step - loss: 0.0032 Epoch 296/300 2/2 [==============================] - 1s 352ms/step - loss: 0.0031 Epoch 297/300 2/2 [==============================] - 1s 351ms/step - loss: 0.0034 Epoch 298/300 2/2 [==============================] - 1s 338ms/step - loss: 0.0039 Epoch 299/300 2/2 [==============================] - 1s 341ms/step - loss: 0.0030 Epoch 300/300 2/2 [==============================] - 1s 364ms/step - loss: 0.0032
The thing with RNNs are that they are limited by vanishing or exploding gradients. This is whey we can use LSTM and/or GRU¶
Build LSTM Model¶
In [41]:
LSTM_model = Sequential()
LSTM_model.add(LSTM(units=300, return_sequences=True, input_shape=(features_set.shape[1], 1)))
LSTM_model.add(Dropout(0.2))
LSTM_model.add(LSTM(units=100, return_sequences=True))
LSTM_model.add(Dropout(0.2))
LSTM_model.add(LSTM(units=100, return_sequences=True))
LSTM_model.add(Dropout(0.2))
LSTM_model.add(LSTM(units=100))
LSTM_model.add(Dropout(0.2))
LSTM_model.add(Dense(units = 1))
Compile LSTM Model using Adam Optimizer¶
In [42]:
LSTM_model.compile(optimizer = 'adam', loss = 'mean_squared_error')
In [43]:
LSTM_model.summary()
Model: "sequential_1"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
lstm (LSTM) (None, 60, 300) 362400
dropout_4 (Dropout) (None, 60, 300) 0
lstm_1 (LSTM) (None, 60, 100) 160400
dropout_5 (Dropout) (None, 60, 100) 0
lstm_2 (LSTM) (None, 60, 100) 80400
dropout_6 (Dropout) (None, 60, 100) 0
lstm_3 (LSTM) (None, 100) 80400
dropout_7 (Dropout) (None, 100) 0
dense_1 (Dense) (None, 1) 101
=================================================================
Total params: 683701 (2.61 MB)
Trainable params: 683701 (2.61 MB)
Non-trainable params: 0 (0.00 Byte)
_________________________________________________________________
Fit LSTM Model¶
In [44]:
LSTM_History = LSTM_model.fit(features_set, labels, epochs = 300, batch_size = 3200)
Epoch 1/300 2/2 [==============================] - 5s 44ms/step - loss: 0.1446 Epoch 2/300 2/2 [==============================] - 0s 36ms/step - loss: 0.0072 Epoch 3/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0060 Epoch 4/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0517 Epoch 5/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0714 Epoch 6/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0684 Epoch 7/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0553 Epoch 8/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0322 Epoch 9/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0137 Epoch 10/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0319 Epoch 11/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0149 Epoch 12/300 2/2 [==============================] - 0s 39ms/step - loss: 0.0085 Epoch 13/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0112 Epoch 14/300 2/2 [==============================] - 0s 40ms/step - loss: 0.0117 Epoch 15/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0068 Epoch 16/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0017 Epoch 17/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0050 Epoch 18/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0049 Epoch 19/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0024 Epoch 20/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0037 Epoch 21/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0036 Epoch 22/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0019 Epoch 23/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0024 Epoch 24/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0021 Epoch 25/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0019 Epoch 26/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0030 Epoch 27/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0025 Epoch 28/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0016 Epoch 29/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0018 Epoch 30/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0016 Epoch 31/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0020 Epoch 32/300 2/2 [==============================] - 0s 41ms/step - loss: 0.0021 Epoch 33/300 2/2 [==============================] - 0s 35ms/step - loss: 0.0015 Epoch 34/300 2/2 [==============================] - 0s 36ms/step - loss: 0.0021 Epoch 35/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0019 Epoch 36/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0016 Epoch 37/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0019 Epoch 38/300 2/2 [==============================] - 0s 39ms/step - loss: 0.0019 Epoch 39/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0015 Epoch 40/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0021 Epoch 41/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0021 Epoch 42/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0015 Epoch 43/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0023 Epoch 44/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0028 Epoch 45/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0019 Epoch 46/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0016 Epoch 47/300 2/2 [==============================] - 0s 44ms/step - loss: 0.0016 Epoch 48/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0014 Epoch 49/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0015 Epoch 50/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0016 Epoch 51/300 2/2 [==============================] - 0s 37ms/step - loss: 0.0014 Epoch 52/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0014 Epoch 53/300 2/2 [==============================] - 0s 35ms/step - loss: 0.0016 Epoch 54/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0028 Epoch 55/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0021 Epoch 56/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0014 Epoch 57/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0026 Epoch 58/300 2/2 [==============================] - 0s 37ms/step - loss: 0.0025 Epoch 59/300 2/2 [==============================] - 0s 64ms/step - loss: 0.0014 Epoch 60/300 2/2 [==============================] - 0s 50ms/step - loss: 0.0018 Epoch 61/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0023 Epoch 62/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0015 Epoch 63/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0013 Epoch 64/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0017 Epoch 65/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0015 Epoch 66/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0013 Epoch 67/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0015 Epoch 68/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0014 Epoch 69/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0015 Epoch 70/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0014 Epoch 71/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0014 Epoch 72/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0013 Epoch 73/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0013 Epoch 74/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0014 Epoch 75/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0014 Epoch 76/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0013 Epoch 77/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0013 Epoch 78/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0013 Epoch 79/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0013 Epoch 80/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0015 Epoch 81/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0013 Epoch 82/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0024 Epoch 83/300 2/2 [==============================] - 0s 37ms/step - loss: 0.0020 Epoch 84/300 2/2 [==============================] - 0s 37ms/step - loss: 0.0013 Epoch 85/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0014 Epoch 86/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0014 Epoch 87/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0021 Epoch 88/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0017 Epoch 89/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0014 Epoch 90/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0017 Epoch 91/300 2/2 [==============================] - 0s 66ms/step - loss: 0.0013 Epoch 92/300 2/2 [==============================] - 0s 69ms/step - loss: 0.0030 Epoch 93/300 2/2 [==============================] - 0s 69ms/step - loss: 0.0027 Epoch 94/300 2/2 [==============================] - 0s 67ms/step - loss: 0.0014 Epoch 95/300 2/2 [==============================] - 0s 75ms/step - loss: 0.0026 Epoch 96/300 2/2 [==============================] - 0s 129ms/step - loss: 0.0027 Epoch 97/300 2/2 [==============================] - 0s 55ms/step - loss: 0.0013 Epoch 98/300 2/2 [==============================] - 0s 51ms/step - loss: 0.0017 Epoch 99/300 2/2 [==============================] - 0s 37ms/step - loss: 0.0013 Epoch 100/300 2/2 [==============================] - 0s 90ms/step - loss: 0.0024 Epoch 101/300 2/2 [==============================] - 0s 40ms/step - loss: 0.0023 Epoch 102/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0013 Epoch 103/300 2/2 [==============================] - 0s 44ms/step - loss: 0.0015 Epoch 104/300 2/2 [==============================] - 0s 37ms/step - loss: 0.0014 Epoch 105/300 2/2 [==============================] - 0s 36ms/step - loss: 0.0013 Epoch 106/300 2/2 [==============================] - 0s 68ms/step - loss: 0.0014 Epoch 107/300 2/2 [==============================] - 0s 67ms/step - loss: 0.0019 Epoch 108/300 2/2 [==============================] - 0s 49ms/step - loss: 0.0017 Epoch 109/300 2/2 [==============================] - 0s 59ms/step - loss: 0.0012 Epoch 110/300 2/2 [==============================] - 0s 68ms/step - loss: 0.0017 Epoch 111/300 2/2 [==============================] - 0s 62ms/step - loss: 0.0014 Epoch 112/300 2/2 [==============================] - 0s 40ms/step - loss: 0.0011 Epoch 113/300 2/2 [==============================] - 0s 85ms/step - loss: 0.0012 Epoch 114/300 2/2 [==============================] - 0s 39ms/step - loss: 0.0014 Epoch 115/300 2/2 [==============================] - 0s 80ms/step - loss: 0.0012 Epoch 116/300 2/2 [==============================] - 0s 66ms/step - loss: 0.0012 Epoch 117/300 2/2 [==============================] - 0s 80ms/step - loss: 0.0011 Epoch 118/300 2/2 [==============================] - 0s 36ms/step - loss: 0.0012 Epoch 119/300 2/2 [==============================] - 0s 59ms/step - loss: 0.0012 Epoch 120/300 2/2 [==============================] - 0s 61ms/step - loss: 0.0012 Epoch 121/300 2/2 [==============================] - 0s 85ms/step - loss: 0.0016 Epoch 122/300 2/2 [==============================] - 0s 76ms/step - loss: 0.0014 Epoch 123/300 2/2 [==============================] - 0s 62ms/step - loss: 0.0012 Epoch 124/300 2/2 [==============================] - 0s 69ms/step - loss: 0.0017 Epoch 125/300 2/2 [==============================] - 0s 76ms/step - loss: 0.0013 Epoch 126/300 2/2 [==============================] - 0s 76ms/step - loss: 0.0026 Epoch 127/300 2/2 [==============================] - 0s 49ms/step - loss: 0.0056 Epoch 128/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0025 Epoch 129/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0018 Epoch 130/300 2/2 [==============================] - 0s 39ms/step - loss: 0.0027 Epoch 131/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0012 Epoch 132/300 2/2 [==============================] - 0s 41ms/step - loss: 0.0022 Epoch 133/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0013 Epoch 134/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0017 Epoch 135/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0019 Epoch 136/300 2/2 [==============================] - 0s 43ms/step - loss: 0.0011 Epoch 137/300 2/2 [==============================] - 0s 36ms/step - loss: 0.0021 Epoch 138/300 2/2 [==============================] - 0s 46ms/step - loss: 0.0023 Epoch 139/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0013 Epoch 140/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0017 Epoch 141/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0018 Epoch 142/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0013 Epoch 143/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0013 Epoch 144/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0016 Epoch 145/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0014 Epoch 146/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0013 Epoch 147/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0015 Epoch 148/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0013 Epoch 149/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0014 Epoch 150/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0012 Epoch 151/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0011 Epoch 152/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0011 Epoch 153/300 2/2 [==============================] - 0s 37ms/step - loss: 0.0013 Epoch 154/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0012 Epoch 155/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0013 Epoch 156/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0012 Epoch 157/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0015 Epoch 158/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0012 Epoch 159/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0017 Epoch 160/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0027 Epoch 161/300 2/2 [==============================] - 0s 46ms/step - loss: 0.0017 Epoch 162/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0016 Epoch 163/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0019 Epoch 164/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0012 Epoch 165/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0027 Epoch 166/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0013 Epoch 167/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0020 Epoch 168/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0024 Epoch 169/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0013 Epoch 170/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0014 Epoch 171/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0014 Epoch 172/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0011 Epoch 173/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0013 Epoch 174/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0015 Epoch 175/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0011 Epoch 176/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0019 Epoch 177/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0019 Epoch 178/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0014 Epoch 179/300 2/2 [==============================] - 0s 76ms/step - loss: 0.0012 Epoch 180/300 2/2 [==============================] - 0s 89ms/step - loss: 0.0012 Epoch 181/300 2/2 [==============================] - 0s 50ms/step - loss: 0.0012 Epoch 182/300 2/2 [==============================] - 0s 61ms/step - loss: 0.0012 Epoch 183/300 2/2 [==============================] - 0s 82ms/step - loss: 0.0012 Epoch 184/300 2/2 [==============================] - 0s 60ms/step - loss: 0.0015 Epoch 185/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0012 Epoch 186/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0014 Epoch 187/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0015 Epoch 188/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0013 Epoch 189/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0013 Epoch 190/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0015 Epoch 191/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0011 Epoch 192/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0014 Epoch 193/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0012 Epoch 194/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0016 Epoch 195/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0022 Epoch 196/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0011 Epoch 197/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0020 Epoch 198/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0013 Epoch 199/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0019 Epoch 200/300 2/2 [==============================] - 0s 48ms/step - loss: 0.0028 Epoch 201/300 2/2 [==============================] - 0s 36ms/step - loss: 0.0013 Epoch 202/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0015 Epoch 203/300 2/2 [==============================] - 0s 54ms/step - loss: 0.0021 Epoch 204/300 2/2 [==============================] - 0s 37ms/step - loss: 0.0013 Epoch 205/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0013 Epoch 206/300 2/2 [==============================] - 0s 43ms/step - loss: 0.0014 Epoch 207/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0013 Epoch 208/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0013 Epoch 209/300 2/2 [==============================] - 0s 45ms/step - loss: 0.0012 Epoch 210/300 2/2 [==============================] - 0s 38ms/step - loss: 0.0012 Epoch 211/300 2/2 [==============================] - 0s 35ms/step - loss: 0.0013 Epoch 212/300 2/2 [==============================] - 0s 101ms/step - loss: 0.0013 Epoch 213/300 2/2 [==============================] - 0s 88ms/step - loss: 0.0012 Epoch 214/300 2/2 [==============================] - 0s 55ms/step - loss: 0.0014 Epoch 215/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0021 Epoch 216/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0012 Epoch 217/300 2/2 [==============================] - 0s 41ms/step - loss: 0.0016 Epoch 218/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0012 Epoch 219/300 2/2 [==============================] - 0s 69ms/step - loss: 0.0012 Epoch 220/300 2/2 [==============================] - 0s 79ms/step - loss: 0.0012 Epoch 221/300 2/2 [==============================] - 0s 44ms/step - loss: 0.0011 Epoch 222/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0014 Epoch 223/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0013 Epoch 224/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0011 Epoch 225/300 2/2 [==============================] - 0s 35ms/step - loss: 0.0013 Epoch 226/300 2/2 [==============================] - 0s 31ms/step - loss: 9.6389e-04 Epoch 227/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0013 Epoch 228/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0019 Epoch 229/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0016 Epoch 230/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0012 Epoch 231/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0022 Epoch 232/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0016 Epoch 233/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0011 Epoch 234/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0013 Epoch 235/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0011 Epoch 236/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0011 Epoch 237/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0011 Epoch 238/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0011 Epoch 239/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0011 Epoch 240/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0010 Epoch 241/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0010 Epoch 242/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0010 Epoch 243/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0012 Epoch 244/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0010 Epoch 245/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0017 Epoch 246/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0025 Epoch 247/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0012 Epoch 248/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0016 Epoch 249/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0013 Epoch 250/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0012 Epoch 251/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0014 Epoch 252/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0012 Epoch 253/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0012 Epoch 254/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0011 Epoch 255/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0011 Epoch 256/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0011 Epoch 257/300 2/2 [==============================] - 0s 37ms/step - loss: 0.0010 Epoch 258/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0011 Epoch 259/300 2/2 [==============================] - 0s 37ms/step - loss: 0.0011 Epoch 260/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0011 Epoch 261/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0011 Epoch 262/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0011 Epoch 263/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0010 Epoch 264/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0010 Epoch 265/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0011 Epoch 266/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0011 Epoch 267/300 2/2 [==============================] - 0s 29ms/step - loss: 9.5714e-04 Epoch 268/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0012 Epoch 269/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0011 Epoch 270/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0010 Epoch 271/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0011 Epoch 272/300 2/2 [==============================] - 0s 36ms/step - loss: 0.0010 Epoch 273/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0011 Epoch 274/300 2/2 [==============================] - 0s 28ms/step - loss: 9.8558e-04 Epoch 275/300 2/2 [==============================] - 0s 30ms/step - loss: 9.8905e-04 Epoch 276/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0010 Epoch 277/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0018 Epoch 278/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0016 Epoch 279/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0012 Epoch 280/300 2/2 [==============================] - 0s 37ms/step - loss: 0.0019 Epoch 281/300 2/2 [==============================] - 0s 41ms/step - loss: 0.0013 Epoch 282/300 2/2 [==============================] - 0s 60ms/step - loss: 0.0011 Epoch 283/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0014 Epoch 284/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0010 Epoch 285/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0010 Epoch 286/300 2/2 [==============================] - 0s 36ms/step - loss: 0.0011 Epoch 287/300 2/2 [==============================] - 0s 33ms/step - loss: 0.0010 Epoch 288/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0010 Epoch 289/300 2/2 [==============================] - 0s 39ms/step - loss: 0.0012 Epoch 290/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0012 Epoch 291/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0012 Epoch 292/300 2/2 [==============================] - 0s 45ms/step - loss: 0.0024 Epoch 293/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0017 Epoch 294/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0010 Epoch 295/300 2/2 [==============================] - 0s 37ms/step - loss: 0.0021 Epoch 296/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0018 Epoch 297/300 2/2 [==============================] - 0s 37ms/step - loss: 0.0011 Epoch 298/300 2/2 [==============================] - 0s 36ms/step - loss: 0.0019 Epoch 299/300 2/2 [==============================] - 0s 46ms/step - loss: 0.0012 Epoch 300/300 2/2 [==============================] - 0s 34ms/step - loss: 0.0013
Build GRU Model¶
In [45]:
GRU_model = Sequential()
GRU_model.add(GRU(units=300, return_sequences=True, input_shape=(features_set.shape[1], 1)))
GRU_model.add(Dropout(0.2))
GRU_model.add(GRU(units=100, return_sequences=True))
GRU_model.add(Dropout(0.2))
GRU_model.add(GRU(units=100, return_sequences=True))
GRU_model.add(Dropout(0.2))
GRU_model.add(GRU(units=100))
GRU_model.add(Dropout(0.2))
GRU_model.add(Dense(units = 1))
Compile GRU Model¶
In [46]:
GRU_model.compile(optimizer = 'adam', loss = 'mean_squared_error')
In [47]:
GRU_model.summary()
Model: "sequential_2"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
gru (GRU) (None, 60, 300) 272700
dropout_8 (Dropout) (None, 60, 300) 0
gru_1 (GRU) (None, 60, 100) 120600
dropout_9 (Dropout) (None, 60, 100) 0
gru_2 (GRU) (None, 60, 100) 60600
dropout_10 (Dropout) (None, 60, 100) 0
gru_3 (GRU) (None, 100) 60600
dropout_11 (Dropout) (None, 100) 0
dense_2 (Dense) (None, 1) 101
=================================================================
Total params: 514601 (1.96 MB)
Trainable params: 514601 (1.96 MB)
Non-trainable params: 0 (0.00 Byte)
_________________________________________________________________
Fit GRU Model¶
In [49]:
GRU_History = GRU_model.fit(features_set, labels, epochs = 300, batch_size = 3200)
Epoch 1/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0015 Epoch 2/300 2/2 [==============================] - 0s 28ms/step - loss: 9.1099e-04 Epoch 3/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0013 Epoch 4/300 2/2 [==============================] - 0s 28ms/step - loss: 8.9488e-04 Epoch 5/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0010 Epoch 6/300 2/2 [==============================] - 0s 25ms/step - loss: 9.8030e-04 Epoch 7/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0016 Epoch 8/300 2/2 [==============================] - 0s 27ms/step - loss: 9.2135e-04 Epoch 9/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0017 Epoch 10/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0013 Epoch 11/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0019 Epoch 12/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0020 Epoch 13/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0013 Epoch 14/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0016 Epoch 15/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0010 Epoch 16/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0019 Epoch 17/300 2/2 [==============================] - 0s 25ms/step - loss: 9.9991e-04 Epoch 18/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0013 Epoch 19/300 2/2 [==============================] - 0s 27ms/step - loss: 7.9811e-04 Epoch 20/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0014 Epoch 21/300 2/2 [==============================] - 0s 27ms/step - loss: 8.3891e-04 Epoch 22/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0020 Epoch 23/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0012 Epoch 24/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0013 Epoch 25/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0010 Epoch 26/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0011 Epoch 27/300 2/2 [==============================] - 0s 25ms/step - loss: 8.1171e-04 Epoch 28/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0018 Epoch 29/300 2/2 [==============================] - 0s 25ms/step - loss: 8.8119e-04 Epoch 30/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0010 Epoch 31/300 2/2 [==============================] - 0s 27ms/step - loss: 9.2756e-04 Epoch 32/300 2/2 [==============================] - 0s 28ms/step - loss: 8.5231e-04 Epoch 33/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0016 Epoch 34/300 2/2 [==============================] - 0s 26ms/step - loss: 8.7137e-04 Epoch 35/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0015 Epoch 36/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0013 Epoch 37/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0012 Epoch 38/300 2/2 [==============================] - 0s 26ms/step - loss: 9.7022e-04 Epoch 39/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0011 Epoch 40/300 2/2 [==============================] - 0s 26ms/step - loss: 8.3904e-04 Epoch 41/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0010 Epoch 42/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0012 Epoch 43/300 2/2 [==============================] - 0s 26ms/step - loss: 8.6205e-04 Epoch 44/300 2/2 [==============================] - 0s 25ms/step - loss: 8.6171e-04 Epoch 45/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0012 Epoch 46/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0012 Epoch 47/300 2/2 [==============================] - 0s 26ms/step - loss: 8.3217e-04 Epoch 48/300 2/2 [==============================] - 0s 25ms/step - loss: 8.6658e-04 Epoch 49/300 2/2 [==============================] - 0s 25ms/step - loss: 8.8774e-04 Epoch 50/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0014 Epoch 51/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0014 Epoch 52/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0010 Epoch 53/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0015 Epoch 54/300 2/2 [==============================] - 0s 28ms/step - loss: 9.4936e-04 Epoch 55/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0026 Epoch 56/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0012 Epoch 57/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0015 Epoch 58/300 2/2 [==============================] - 0s 25ms/step - loss: 8.3054e-04 Epoch 59/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0015 Epoch 60/300 2/2 [==============================] - 0s 26ms/step - loss: 8.6574e-04 Epoch 61/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0013 Epoch 62/300 2/2 [==============================] - 0s 27ms/step - loss: 9.6525e-04 Epoch 63/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0014 Epoch 64/300 2/2 [==============================] - 0s 25ms/step - loss: 8.9004e-04 Epoch 65/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0010 Epoch 66/300 2/2 [==============================] - 0s 25ms/step - loss: 8.4782e-04 Epoch 67/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0013 Epoch 68/300 2/2 [==============================] - 0s 24ms/step - loss: 9.2502e-04 Epoch 69/300 2/2 [==============================] - 0s 25ms/step - loss: 8.1751e-04 Epoch 70/300 2/2 [==============================] - 0s 27ms/step - loss: 8.3758e-04 Epoch 71/300 2/2 [==============================] - 0s 27ms/step - loss: 9.8058e-04 Epoch 72/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0010 Epoch 73/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0011 Epoch 74/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0023 Epoch 75/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0013 Epoch 76/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0031 Epoch 77/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0011 Epoch 78/300 2/2 [==============================] - 0s 31ms/step - loss: 0.0021 Epoch 79/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0012 Epoch 80/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0013 Epoch 81/300 2/2 [==============================] - 0s 29ms/step - loss: 8.8027e-04 Epoch 82/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0010 Epoch 83/300 2/2 [==============================] - 0s 24ms/step - loss: 8.8506e-04 Epoch 84/300 2/2 [==============================] - 0s 25ms/step - loss: 7.2732e-04 Epoch 85/300 2/2 [==============================] - 0s 25ms/step - loss: 9.6369e-04 Epoch 86/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0010 Epoch 87/300 2/2 [==============================] - 0s 27ms/step - loss: 8.2092e-04 Epoch 88/300 2/2 [==============================] - 0s 24ms/step - loss: 9.1392e-04 Epoch 89/300 2/2 [==============================] - 0s 28ms/step - loss: 8.8695e-04 Epoch 90/300 2/2 [==============================] - 0s 37ms/step - loss: 8.8345e-04 Epoch 91/300 2/2 [==============================] - 0s 26ms/step - loss: 9.1133e-04 Epoch 92/300 2/2 [==============================] - 0s 27ms/step - loss: 8.7299e-04 Epoch 93/300 2/2 [==============================] - 0s 26ms/step - loss: 9.2683e-04 Epoch 94/300 2/2 [==============================] - 0s 26ms/step - loss: 7.7390e-04 Epoch 95/300 2/2 [==============================] - 0s 28ms/step - loss: 7.6670e-04 Epoch 96/300 2/2 [==============================] - 0s 25ms/step - loss: 8.1387e-04 Epoch 97/300 2/2 [==============================] - 0s 42ms/step - loss: 8.1719e-04 Epoch 98/300 2/2 [==============================] - 0s 45ms/step - loss: 9.2919e-04 Epoch 99/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0011 Epoch 100/300 2/2 [==============================] - 0s 30ms/step - loss: 8.4066e-04 Epoch 101/300 2/2 [==============================] - 0s 32ms/step - loss: 7.7987e-04 Epoch 102/300 2/2 [==============================] - 0s 29ms/step - loss: 7.5439e-04 Epoch 103/300 2/2 [==============================] - 0s 29ms/step - loss: 7.8905e-04 Epoch 104/300 2/2 [==============================] - 0s 26ms/step - loss: 8.0700e-04 Epoch 105/300 2/2 [==============================] - 0s 27ms/step - loss: 8.4674e-04 Epoch 106/300 2/2 [==============================] - 0s 25ms/step - loss: 7.9269e-04 Epoch 107/300 2/2 [==============================] - 0s 23ms/step - loss: 7.6097e-04 Epoch 108/300 2/2 [==============================] - 0s 27ms/step - loss: 9.7291e-04 Epoch 109/300 2/2 [==============================] - 0s 26ms/step - loss: 8.9301e-04 Epoch 110/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0012 Epoch 111/300 2/2 [==============================] - 0s 28ms/step - loss: 9.4533e-04 Epoch 112/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0020 Epoch 113/300 2/2 [==============================] - 0s 25ms/step - loss: 9.9106e-04 Epoch 114/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0011 Epoch 115/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0012 Epoch 116/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0011 Epoch 117/300 2/2 [==============================] - 0s 26ms/step - loss: 9.8360e-04 Epoch 118/300 2/2 [==============================] - 0s 79ms/step - loss: 8.8835e-04 Epoch 119/300 2/2 [==============================] - 0s 40ms/step - loss: 0.0011 Epoch 120/300 2/2 [==============================] - 0s 28ms/step - loss: 9.1511e-04 Epoch 121/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0011 Epoch 122/300 2/2 [==============================] - 0s 26ms/step - loss: 9.0594e-04 Epoch 123/300 2/2 [==============================] - 0s 27ms/step - loss: 8.9259e-04 Epoch 124/300 2/2 [==============================] - 0s 26ms/step - loss: 8.9441e-04 Epoch 125/300 2/2 [==============================] - 0s 28ms/step - loss: 9.4429e-04 Epoch 126/300 2/2 [==============================] - 0s 26ms/step - loss: 8.8247e-04 Epoch 127/300 2/2 [==============================] - 0s 26ms/step - loss: 9.8742e-04 Epoch 128/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0011 Epoch 129/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0040 Epoch 130/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0013 Epoch 131/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0013 Epoch 132/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0013 Epoch 133/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0012 Epoch 134/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0015 Epoch 135/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0018 Epoch 136/300 2/2 [==============================] - 0s 26ms/step - loss: 7.9454e-04 Epoch 137/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0014 Epoch 138/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0012 Epoch 139/300 2/2 [==============================] - 0s 28ms/step - loss: 7.4842e-04 Epoch 140/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0011 Epoch 141/300 2/2 [==============================] - 0s 26ms/step - loss: 9.3591e-04 Epoch 142/300 2/2 [==============================] - 0s 26ms/step - loss: 7.8746e-04 Epoch 143/300 2/2 [==============================] - 0s 27ms/step - loss: 8.3400e-04 Epoch 144/300 2/2 [==============================] - 0s 25ms/step - loss: 7.6791e-04 Epoch 145/300 2/2 [==============================] - 0s 25ms/step - loss: 9.7392e-04 Epoch 146/300 2/2 [==============================] - 0s 26ms/step - loss: 7.7502e-04 Epoch 147/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0012 Epoch 148/300 2/2 [==============================] - 0s 24ms/step - loss: 9.0636e-04 Epoch 149/300 2/2 [==============================] - 0s 29ms/step - loss: 9.2718e-04 Epoch 150/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0011 Epoch 151/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0013 Epoch 152/300 2/2 [==============================] - 0s 28ms/step - loss: 8.5817e-04 Epoch 153/300 2/2 [==============================] - 0s 25ms/step - loss: 8.6383e-04 Epoch 154/300 2/2 [==============================] - 0s 28ms/step - loss: 9.2905e-04 Epoch 155/300 2/2 [==============================] - 0s 26ms/step - loss: 8.5568e-04 Epoch 156/300 2/2 [==============================] - 0s 26ms/step - loss: 7.1449e-04 Epoch 157/300 2/2 [==============================] - 0s 25ms/step - loss: 8.4387e-04 Epoch 158/300 2/2 [==============================] - 0s 25ms/step - loss: 9.0697e-04 Epoch 159/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0010 Epoch 160/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0013 Epoch 161/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0018 Epoch 162/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0012 Epoch 163/300 2/2 [==============================] - 0s 30ms/step - loss: 0.0016 Epoch 164/300 2/2 [==============================] - 0s 32ms/step - loss: 0.0011 Epoch 165/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0015 Epoch 166/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0012 Epoch 167/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0019 Epoch 168/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0011 Epoch 169/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0018 Epoch 170/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0012 Epoch 171/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0013 Epoch 172/300 2/2 [==============================] - 0s 26ms/step - loss: 9.2894e-04 Epoch 173/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0010 Epoch 174/300 2/2 [==============================] - 0s 26ms/step - loss: 9.5806e-04 Epoch 175/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0012 Epoch 176/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0017 Epoch 177/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0011 Epoch 178/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0011 Epoch 179/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0011 Epoch 180/300 2/2 [==============================] - 0s 25ms/step - loss: 8.3284e-04 Epoch 181/300 2/2 [==============================] - 0s 25ms/step - loss: 8.5115e-04 Epoch 182/300 2/2 [==============================] - 0s 27ms/step - loss: 8.2456e-04 Epoch 183/300 2/2 [==============================] - 0s 25ms/step - loss: 9.9511e-04 Epoch 184/300 2/2 [==============================] - 0s 25ms/step - loss: 8.9341e-04 Epoch 185/300 2/2 [==============================] - 0s 26ms/step - loss: 8.9302e-04 Epoch 186/300 2/2 [==============================] - 0s 27ms/step - loss: 8.2447e-04 Epoch 187/300 2/2 [==============================] - 0s 29ms/step - loss: 7.4186e-04 Epoch 188/300 2/2 [==============================] - 0s 26ms/step - loss: 8.4731e-04 Epoch 189/300 2/2 [==============================] - 0s 25ms/step - loss: 8.7227e-04 Epoch 190/300 2/2 [==============================] - 0s 25ms/step - loss: 6.6852e-04 Epoch 191/300 2/2 [==============================] - 0s 25ms/step - loss: 7.4747e-04 Epoch 192/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0013 Epoch 193/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0011 Epoch 194/300 2/2 [==============================] - 0s 24ms/step - loss: 9.1299e-04 Epoch 195/300 2/2 [==============================] - 0s 26ms/step - loss: 7.6903e-04 Epoch 196/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0014 Epoch 197/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0013 Epoch 198/300 2/2 [==============================] - 0s 27ms/step - loss: 9.2971e-04 Epoch 199/300 2/2 [==============================] - 0s 24ms/step - loss: 9.1601e-04 Epoch 200/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0012 Epoch 201/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0010 Epoch 202/300 2/2 [==============================] - 0s 26ms/step - loss: 7.9099e-04 Epoch 203/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0011 Epoch 204/300 2/2 [==============================] - 0s 27ms/step - loss: 8.9596e-04 Epoch 205/300 2/2 [==============================] - 0s 26ms/step - loss: 7.2944e-04 Epoch 206/300 2/2 [==============================] - 0s 28ms/step - loss: 7.7591e-04 Epoch 207/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0011 Epoch 208/300 2/2 [==============================] - 0s 27ms/step - loss: 7.6260e-04 Epoch 209/300 2/2 [==============================] - 0s 25ms/step - loss: 9.1872e-04 Epoch 210/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0011 Epoch 211/300 2/2 [==============================] - 0s 27ms/step - loss: 8.5457e-04 Epoch 212/300 2/2 [==============================] - 0s 26ms/step - loss: 8.6778e-04 Epoch 213/300 2/2 [==============================] - 0s 25ms/step - loss: 8.9113e-04 Epoch 214/300 2/2 [==============================] - 0s 25ms/step - loss: 7.8802e-04 Epoch 215/300 2/2 [==============================] - 0s 28ms/step - loss: 8.8206e-04 Epoch 216/300 2/2 [==============================] - 0s 29ms/step - loss: 9.7884e-04 Epoch 217/300 2/2 [==============================] - 0s 28ms/step - loss: 7.7089e-04 Epoch 218/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0011 Epoch 219/300 2/2 [==============================] - 0s 31ms/step - loss: 9.4775e-04 Epoch 220/300 2/2 [==============================] - 0s 26ms/step - loss: 9.0550e-04 Epoch 221/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0010 Epoch 222/300 2/2 [==============================] - 0s 25ms/step - loss: 8.9265e-04 Epoch 223/300 2/2 [==============================] - 0s 25ms/step - loss: 8.9822e-04 Epoch 224/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0013 Epoch 225/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0015 Epoch 226/300 2/2 [==============================] - 0s 26ms/step - loss: 9.6624e-04 Epoch 227/300 2/2 [==============================] - 0s 24ms/step - loss: 8.3240e-04 Epoch 228/300 2/2 [==============================] - 0s 29ms/step - loss: 9.8412e-04 Epoch 229/300 2/2 [==============================] - 0s 25ms/step - loss: 8.1253e-04 Epoch 230/300 2/2 [==============================] - 0s 25ms/step - loss: 7.0318e-04 Epoch 231/300 2/2 [==============================] - 0s 24ms/step - loss: 8.4021e-04 Epoch 232/300 2/2 [==============================] - 0s 28ms/step - loss: 7.3491e-04 Epoch 233/300 2/2 [==============================] - 0s 27ms/step - loss: 7.1771e-04 Epoch 234/300 2/2 [==============================] - 0s 25ms/step - loss: 8.4999e-04 Epoch 235/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0015 Epoch 236/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0011 Epoch 237/300 2/2 [==============================] - 0s 25ms/step - loss: 7.9317e-04 Epoch 238/300 2/2 [==============================] - 0s 28ms/step - loss: 9.1153e-04 Epoch 239/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0010 Epoch 240/300 2/2 [==============================] - 0s 27ms/step - loss: 7.6901e-04 Epoch 241/300 2/2 [==============================] - 0s 26ms/step - loss: 9.5661e-04 Epoch 242/300 2/2 [==============================] - 0s 28ms/step - loss: 8.7541e-04 Epoch 243/300 2/2 [==============================] - 0s 27ms/step - loss: 9.6288e-04 Epoch 244/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0011 Epoch 245/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0011 Epoch 246/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0015 Epoch 247/300 2/2 [==============================] - 0s 47ms/step - loss: 0.0011 Epoch 248/300 2/2 [==============================] - 0s 44ms/step - loss: 0.0010 Epoch 249/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0012 Epoch 250/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0014 Epoch 251/300 2/2 [==============================] - 0s 24ms/step - loss: 8.9917e-04 Epoch 252/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0015 Epoch 253/300 2/2 [==============================] - 0s 26ms/step - loss: 6.7499e-04 Epoch 254/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0025 Epoch 255/300 2/2 [==============================] - 0s 56ms/step - loss: 0.0012 Epoch 256/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0015 Epoch 257/300 2/2 [==============================] - 0s 28ms/step - loss: 8.4359e-04 Epoch 258/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0018 Epoch 259/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0014 Epoch 260/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0011 Epoch 261/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0013 Epoch 262/300 2/2 [==============================] - 0s 24ms/step - loss: 0.0014 Epoch 263/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0013 Epoch 264/300 2/2 [==============================] - 0s 24ms/step - loss: 8.6234e-04 Epoch 265/300 2/2 [==============================] - 0s 29ms/step - loss: 0.0019 Epoch 266/300 2/2 [==============================] - 0s 27ms/step - loss: 8.8534e-04 Epoch 267/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0018 Epoch 268/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0021 Epoch 269/300 2/2 [==============================] - 0s 28ms/step - loss: 0.0010 Epoch 270/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0020 Epoch 271/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0012 Epoch 272/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0015 Epoch 273/300 2/2 [==============================] - 0s 25ms/step - loss: 9.8750e-04 Epoch 274/300 2/2 [==============================] - 0s 26ms/step - loss: 9.6379e-04 Epoch 275/300 2/2 [==============================] - 0s 26ms/step - loss: 7.9564e-04 Epoch 276/300 2/2 [==============================] - 0s 27ms/step - loss: 7.8559e-04 Epoch 277/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0016 Epoch 278/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0012 Epoch 279/300 2/2 [==============================] - 0s 27ms/step - loss: 8.1380e-04 Epoch 280/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0012 Epoch 281/300 2/2 [==============================] - 0s 27ms/step - loss: 8.8170e-04 Epoch 282/300 2/2 [==============================] - 0s 26ms/step - loss: 9.4164e-04 Epoch 283/300 2/2 [==============================] - 0s 27ms/step - loss: 0.0012 Epoch 284/300 2/2 [==============================] - 0s 24ms/step - loss: 8.7087e-04 Epoch 285/300 2/2 [==============================] - 0s 25ms/step - loss: 7.6947e-04 Epoch 286/300 2/2 [==============================] - 0s 30ms/step - loss: 8.2362e-04 Epoch 287/300 2/2 [==============================] - 0s 26ms/step - loss: 8.8744e-04 Epoch 288/300 2/2 [==============================] - 0s 32ms/step - loss: 7.4360e-04 Epoch 289/300 2/2 [==============================] - 0s 24ms/step - loss: 7.9196e-04 Epoch 290/300 2/2 [==============================] - 0s 25ms/step - loss: 6.8367e-04 Epoch 291/300 2/2 [==============================] - 0s 24ms/step - loss: 8.5388e-04 Epoch 292/300 2/2 [==============================] - 0s 29ms/step - loss: 7.4762e-04 Epoch 293/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0011 Epoch 294/300 2/2 [==============================] - 0s 27ms/step - loss: 9.3496e-04 Epoch 295/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0010 Epoch 296/300 2/2 [==============================] - 0s 24ms/step - loss: 6.7403e-04 Epoch 297/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0023 Epoch 298/300 2/2 [==============================] - 0s 26ms/step - loss: 0.0017 Epoch 299/300 2/2 [==============================] - 0s 25ms/step - loss: 0.0013 Epoch 300/300 2/2 [==============================] - 0s 25ms/step - loss: 9.3637e-04
Evaluate The Model¶
Evaluate Model Training¶
In [50]:
fig = go.Figure()
# Training loss trace
fig.add_trace(go.Scatter(x=list(range(1, len(RNN_History.history['loss']) + 1)),
y=RNN_History.history['loss'],
mode='lines',
name='RNN Training Loss'))
fig.add_trace(go.Scatter(x=list(range(1, len(LSTM_History.history['loss']) + 1)),
y=LSTM_History.history['loss'],
mode='lines',
name='LSTM Training Loss'))
fig.add_trace(go.Scatter(x=list(range(1, len(GRU_History.history['loss']) + 1)),
y=GRU_History.history['loss'],
mode='lines',
name='GRU Training Loss'))
fig.update_layout(title='Training and Validation Loss',
xaxis_title='Epoch',
yaxis_title='Loss',
legend_title='Loss Type',
hovermode='x',
hoverlabel=dict(bgcolor='white', font_size=12, font_family='Rockwell'),
template='plotly_white')
fig.show()
In [57]:
google_testing_complete = pd.read_csv("./Google_Stock_Test (2023).csv")
google_testing_processed = google_testing_complete.iloc[:, 1:2].values
google_total = pd.concat((google_training_complete['Close'], google_testing_complete['Close']), axis=0)
test_inputs = google_total[len(google_total) - len(google_testing_complete) - 60:].values
test_inputs
Out[57]:
array([101.419998, 98.68 , 97.860001, 97.18 , 97.559998,
99.059998, 96.559998, 99.970001, 100.769997, 99.629997,
99.970001, 101.129997, 102.519997, 104.480003, 94.93 ,
92.220001, 96.290001, 94.510002, 90.470001, 86.970001,
83.43 , 86.580002, 88.489998, 88.900002, 87.32 ,
93.940002, 96.410004, 95.699997, 98.440002, 98.849998,
98.360001, 97.43 , 95.599998, 97.050003, 98.459999,
97.459999, 96.050003, 95.190002, 100.989998, 100.989998,
100.440002, 99.480003, 96.980003, 94.940002, 93.709999,
92.830002, 93.309998, 95.629997, 95.07 , 90.860001,
90.260002, 88.440002, 89.019997, 89.580002, 87.760002,
89.230003, 87.389999, 86.019997, 88.449997, 88.230003,
89.120003, 88.080002, 86.199997, 87.339996, 88.019997,
88.419998, 91.519997, 91.129997, 92.120003, 91.290001,
91.120003, 93.050003, 98.019997, 99.790001, 97.699997,
95.220001, 97.519997, 99.370003, 96.940002, 98.839996,
100.43 , 107.739998, 104.779999, 102.900002, 107.639999,
99.370003, 95.010002, 94.57 , 94.610001, 94.68 ,
96.940002, 95.510002, 94.349998, 91.790001, 91.650002,
90.889999, 89.129997, 89.870003, 90.059998, 90.360001,
92. , 93.650002, 95.129997, 93.860001, 94.25 ,
92.32 , 90.629997, 91.110001, 93.970001, 96.110001,
100.32 , 101.620003, 101.220001, 104.919998, 103.370003,
105.599998, 105.440002, 102.459999, 101.029999, 101.389999,
100.889999, 103.730003, 104.360001, 104.720001, 104.470001,
108.419998, 106.440002, 105.349998, 104.639999, 107.43 ,
108.870003, 105.970001, 104.5 , 104.18 , 105.290001,
105.410004, 105.970001, 103.849998, 103.709999, 107.589996,
107.339996, 107.199997, 105.32 , 105.410004, 104.690002,
105.57 , 107.769997, 107.349998, 111.75 , 116.57 ,
117.510002, 116.510002, 119.510002, 120.839996, 122.830002,
122.760002, 125.050003, 122.559998, 120.900002, 123.480003,
124.610001, 123.669998, 122.870003, 123.720001, 124.669998,
126.010002, 127.309998, 122.5 , 122.139999, 122.230003,
123.639999, 123.830002, 123.669998, 125.089996, 123.529999,
123.099998, 120.550003, 123.150002, 122.339996, 118.339996,
118.330002, 120.18 , 119.099998, 119.699997, 119.900002,
121.75 , 120.110001, 119.480003, 116.449997, 117.139999,
118.93 , 124.540001, 125.419998, 124.650002, 123.760002,
122.029999, 119.199997, 120.019997, 121.529999, 122.209999,
129.270004, 129.399994, 132.580002])
In [58]:
test_inputs = test_inputs.reshape(-1,1)
test_inputs = scaler.transform(test_inputs)
test_features = []
for i in range(60, len(test_inputs)):
test_features.append(test_inputs[i-60:i, 0])
test_features = np.array(test_features)
test_features = np.reshape(test_features, (test_features.shape[0], test_features.shape[1], 1))
test_features.shape
Out[58]:
(143, 60, 1)
In [59]:
# Test the model on the test set
## RNN
RNN_test_loss = RNN_model.evaluate(test_features, google_testing_processed)
print(f'RNN Test Loss: {RNN_test_loss}')
## LSTM
LSTM_test_loss = LSTM_model.evaluate(test_features, google_testing_processed)
print(f'LSTM Test Loss: {LSTM_test_loss}')
## GRU
GRU_test_loss = GRU_model.evaluate(test_features, google_testing_processed)
print(f'LSTM Test Loss: {GRU_test_loss}')
5/5 [==============================] - 1s 66ms/step - loss: 11545.4990 RNN Test Loss: 11545.4990234375 5/5 [==============================] - 1s 16ms/step - loss: 11550.0293 LSTM Test Loss: 11550.029296875 5/5 [==============================] - 1s 18ms/step - loss: 11575.7695 LSTM Test Loss: 11575.76953125
Forecast Markets¶
In [60]:
# RNN
RNN_predictions = RNN_model.predict(test_features)
RNN_predictions = scaler.inverse_transform(RNN_predictions)
# LSTM
LSTM_predictions = LSTM_model.predict(test_features)
LSTM_predictions = scaler.inverse_transform(LSTM_predictions)
# GRU
GRU_predictions = GRU_model.predict(test_features)
GRU_predictions = scaler.inverse_transform(GRU_predictions)
5/5 [==============================] - 1s 51ms/step 5/5 [==============================] - 1s 14ms/step 5/5 [==============================] - 1s 10ms/step
In [61]:
google_testing_complete
Out[61]:
| Date | Open | High | Low | Close | Adj Close | Volume | |
|---|---|---|---|---|---|---|---|
| 0 | 2023-01-03 | 89.589996 | 91.050003 | 88.519997 | 89.120003 | 89.120003 | 28131200 |
| 1 | 2023-01-04 | 90.349998 | 90.650002 | 87.269997 | 88.080002 | 88.080002 | 34854800 |
| 2 | 2023-01-05 | 87.470001 | 87.570000 | 85.900002 | 86.199997 | 86.199997 | 27194400 |
| 3 | 2023-01-06 | 86.790001 | 87.690002 | 84.860001 | 87.339996 | 87.339996 | 41381500 |
| 4 | 2023-01-09 | 88.360001 | 90.050003 | 87.860001 | 88.019997 | 88.019997 | 29003900 |
| ... | ... | ... | ... | ... | ... | ... | ... |
| 138 | 2023-07-24 | 121.660004 | 123.000000 | 120.980003 | 121.529999 | 121.529999 | 29686100 |
| 139 | 2023-07-25 | 121.360001 | 123.150002 | 121.019997 | 122.209999 | 122.209999 | 52509600 |
| 140 | 2023-07-26 | 130.070007 | 130.979996 | 128.320007 | 129.270004 | 129.270004 | 61682100 |
| 141 | 2023-07-27 | 131.669998 | 133.240005 | 128.789993 | 129.399994 | 129.399994 | 44952100 |
| 142 | 2023-07-28 | 130.779999 | 133.740005 | 130.570007 | 132.580002 | 132.580002 | 36572900 |
143 rows × 7 columns
In [62]:
google_prediction = google_testing_complete['Date']
google_prediction = pd.DataFrame(google_prediction)
google_prediction['RNN_Open'] = RNN_predictions
google_prediction['LSTM_Open'] = LSTM_predictions
google_prediction['GRU_Open'] = GRU_predictions
google_prediction
Out[62]:
| Date | RNN_Open | LSTM_Open | GRU_Open | |
|---|---|---|---|---|
| 0 | 2023-01-03 | 89.519333 | 98.401093 | 78.831085 |
| 1 | 2023-01-04 | 89.489113 | 97.782120 | 79.041344 |
| 2 | 2023-01-05 | 89.636574 | 97.219818 | 79.156418 |
| 3 | 2023-01-06 | 89.770462 | 96.714958 | 78.887222 |
| 4 | 2023-01-09 | 89.799355 | 96.264481 | 78.604805 |
| ... | ... | ... | ... | ... |
| 138 | 2023-07-24 | 136.945770 | 127.663452 | 108.560730 |
| 139 | 2023-07-25 | 136.781494 | 127.848602 | 108.264091 |
| 140 | 2023-07-26 | 136.534897 | 128.019958 | 108.350693 |
| 141 | 2023-07-27 | 136.514725 | 128.205750 | 109.546196 |
| 142 | 2023-07-28 | 136.901825 | 128.438797 | 111.187195 |
143 rows × 4 columns
In [64]:
fig = go.Figure()
# Actual Apple stock price trace
fig.add_trace(go.Scatter(x=google_testing_complete.Date,
y=google_testing_complete.Open,
mode='lines',
name='Actual Google Stock Price',
line=dict(color='blue')))
# RNN Predicted Apple stock price trace
fig.add_trace(go.Scatter(x=google_prediction.Date,
y=google_prediction.RNN_Open,
mode='lines',
name='RNN Predicted Google Stock Price',
line=dict(color='red')))
# LSTM Predicted Apple stock price trace
fig.add_trace(go.Scatter(x=google_prediction.Date,
y=google_prediction.LSTM_Open,
mode='lines',
name='LSTM Predicted Google Stock Price',
line=dict(color='green')))
# LSTM Predicted Apple stock price trace
fig.add_trace(go.Scatter(x=google_prediction.Date,
y=google_prediction.GRU_Open,
mode='lines',
name='GRU Predicted Google Stock Price',
line=dict(color='yellow')))
fig.update_layout(title='Google Stock Price Forecast',
xaxis_title='Date',
yaxis_title='Google Stock Price',
legend_title='Price Type',
hovermode='x',
hoverlabel=dict(bgcolor='white', font_size=12, font_family='Rockwell'),
template='plotly_white')
fig.show()
In [ ]: